Re: [PATCH 02/11] mm,migration: Do not try to migrate unmapped anonymous pages

[Minchan Kim]

On Wed, Mar 17, 2010 at 11:12 AM, KAMEZAWA Hiroyuki
> BTW, I doubt freeing anon_vma can happen even when we check mapcount.
>
> "unmap" is 2-stage operation.
> Â Â Â Â1. unmap_vmas() => modify ptes, free pages, etc.
> Â Â Â Â2. free_pgtables() => free pgtables, unlink vma and free it.
>
> Then, if migration is enough slow.
>
> Â Â Â ÂMigration(): Â Â Â Â Â Â Â Â Â Â Â Â Â ÂExit():
> Â Â Â Âcheck mapcount
> Â Â Â Ârcu_read_lock
> Â Â Â Âpte_lock
> Â Â Â Âreplace pte with migration pte
> Â Â Â Âpte_unlock
> Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Âpte_lock
> Â Â Â Âcopy page etc... Â Â Â Â Â Â Â Â Â Â Â Âzap pte (clear pte)
> Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Âpte_unlock
> Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Âfree_pgtables
> Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â->free vma
> Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â->free anon_vma
> Â Â Â Âpte_lock
> Â Â Â Âremap pte with new pfn(fail)
> Â Â Â Âpte_unlock
>
>    Âlock anon_vma->lock       # modification after free.
> Â Â Â Âcheck list is empty

check list is empty?
Do you mean anon_vma->head?

If it is, is it possible that that list isn't empty since anon_vma is
used by others due to
SLAB_DESTROY_BY_RCU?

but such case is handled by page_check_address, vma_address, I think.

> Â Â Â Âunlock anon_vma->lock
> Â Â Â Âfree anon_vma
> Â Â Â Ârcu_read_unlock
>
>
> Hmm. IIUC, anon_vma is allocated as SLAB_DESTROY_BY_RCU. Then, while
> rcu_read_lock() is taken, anon_vma is anon_vma even if freed. But it
> may reused as anon_vma for someone else.
> (IOW, it may be reused but never pushed back to general purpose memory
> Âuntil RCU grace period.)
> Then, touching anon_vma->lock never cause any corruption.
>
> Does use-after-free check for SLAB_DESTROY_BY_RCU correct behavior ?

Could you elaborate your point?

> Above case is not use-after-free. It's safe and expected sequence.
>
> Thanks,
> -Kame
>
>
>
>> > ---
>> > Âmm/migrate.c | Â 13 +++++++++++++
>> > Â1 files changed, 13 insertions(+), 0 deletions(-)
>> >
>> > diff --git a/mm/migrate.c b/mm/migrate.c
>> > index 98eaaf2..6eb1efe 100644
>> > --- a/mm/migrate.c
>> > +++ b/mm/migrate.c
>> > @@ -603,6 +603,19 @@ static int unmap_and_move(new_page_t get_new_page, unsigned long private,
>> > Â Â Â*/
>> > Â Â if (PageAnon(page)) {
>> > Â Â Â Â Â Â rcu_read_lock();
>> > +
>> > + Â Â Â Â Â /*
>> > + Â Â Â Â Â Â* If the page has no mappings any more, just bail. An
>> > + Â Â Â Â Â Â* unmapped anon page is likely to be freed soon but worse,
>> > + Â Â Â Â Â Â* it's possible its anon_vma disappeared between when
>> > + Â Â Â Â Â Â* the page was isolated and when we reached here while
>> > + Â Â Â Â Â Â* the RCU lock was not held
>> > + Â Â Â Â Â Â*/
>> > + Â Â Â Â Â if (!page_mapcount(page)) {
>> > + Â Â Â Â Â Â Â Â Â rcu_read_unlock();
>> > + Â Â Â Â Â Â Â Â Â goto uncharge;
>> > + Â Â Â Â Â }
>> > +
>> > Â Â Â Â Â Â rcu_locked = 1;
>> > Â Â Â Â Â Â anon_vma = page_anon_vma(page);
>> > Â Â Â Â Â Â atomic_inc(&anon_vma->migrate_refcount);
>> >
>>
>> --
>> To unsubscribe, send a message with 'unsubscribe linux-mm' in
>> the body to majordomo@xxxxxxxxxx ÂFor more info on Linux MM,
>> see: http://www.linux-mm.org/ .
>> Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>
>>
>
>



-- 
Kind regards,
Minchan Kim
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/